RecGPT: Pre-Training and Evaluation Data for Sequential Recommendation

RecGPT-Dataset contains the pre-training, evaluation, and test data used in the paper 'RecGPT: A Foundation Model for Sequential Recommendation'. The dataset was created by the HKUDS organization and last updated on June 12, 2025. It serves as the core data for training and evaluating the RecGPT foundation model.

Use Cases

• Pre-training foundation models for sequential recommendation based on the described pre-training data.
• Evaluating sequential recommendation algorithms based on the described evaluation data.
• Benchmarking new sequential recommendation models against the RecGPT test data.
• Studying the transfer learning capabilities of foundation models in recommendation tasks based on the dataset's structure.

Strengths

• Dataset is directly associated with a published research paper, providing academic context.
• Data is structured for a specific purpose: pre-training, evaluation, and testing of a foundation model.
• Last update timestamp (2025-06-12 15:17:31) indicates recent maintenance.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

hkuds (Hong Kong University Data Science group, inferred).

Collection Method

Created for the RecGPT research project; specific collection method is not detailed.

Freshness

Last updated 2025-06-12 15:17:31.